Abstract
The cellular prion protein PrPC confers susceptibility to transmissible spongiform encephalopathies, yet its normal function is unknown. Although PrPC-deficient mice develop and live normally, expression of amino proximally truncated PrPC (ΔPrP) or of its structural homolog Doppel (Dpl) causes cerebellar degeneration that is prevented by coexpression of full-length PrPC. We now report that mice expressing ΔPrP or Dpl suffer from widespread leukoencephalopathy. Oligodendrocyte-specific expression of full-length PrPC under control of the myelin basic protein (MBP) promoter repressed leukoencephalopathy and vastly extended survival but did not prevent cerebellar granule cell (CGC) degeneration. Conversely, neuron-specific PrPC expression under control of the neuron-specific enolase (NSE) promoter antagonized CGC degeneration but not leukoencephalopathy. PrPC was found in purified myelin and in cultured oligodendrocytes of both wild-type and MBP-PrP transgenic mice but not in NSE-PrP mice. These results identify white-matter damage as an extraneuronal PrP-associated pathology and suggest a previously unrecognized role of PrPC in myelin maintenance.
Introduction
The infectious agent causing transmissible spongiform encephalopathies (TSE) was termed prion (Prusiner, 1982). Its only known constituent is PrPSc, a TSE-associated isoform of the cellular prion protein PrPC. Ablation of the Prnp gene, which encodes PrPC, abrogates prion replication (Büeler et al., 1993) and scrapie pathogenesis (Brandner et al., 1996), but the physiological function of PrPC has remained elusive (Aguzzi and Polymenidou, 2004). Prnpo/o mice show no obvious developmental defects and enjoy a normal life expectancy (Büeler et al., 1992). Subtle changes in circadian rhythms (Tobler et al., 1996), alterations of hippocampal function (Collinge et al., 1994; Colling et al., 1997; Herms et al., 1999; Mallucci et al., 2002), and behavioral abnormalities have been described. However, some of these phenotypes were not reproduced by others (Lledo et al., 1996), and none were clarified in molecular terms. PrPC binds copper (Brown et al., 1997; Kretzschmar et al., 2000) and may have antiapoptotic properties (Kuwahara et al., 1999; Bounhar et al., 2001; Chiarini et al., 2002; Zanata et al., 2002) but could also sensitize neurons to apoptotic stimuli (Paitel et al., 2003). A possible dismutase activity of PrPC was not confirmed in vivo (Waggoner et al., 2000; Hutter et al., 2003). Several factors bind PrPC, including the laminin receptor precursor protein (Rieger et al., 1997), heparan sulfate (Caughey et al., 1994), neural cell adhesion molecule (NCAM) (Schmitt-Ulms et al., 2001), and bcl-2 (Kurschner et al., 1995). However, none of these interactions were shown to be functionally significant. Hence, no unified view of PrPC function has emerged from the above observations.
Transgenic expression of amino proximally truncated PrPC mutants causes early-onset ataxia associated with progressive cerebellar granule cell (CGC) degeneration and death at 3-4 months of age, also referred to as Shmerling's syndrome (Shmerling et al., 1998). Only truncated versions of PrPC encompassing half (amino acids Δ34-121) or all (amino acids Δ34-134) of the highly conserved hydrophobic transmembrane domain 1 (TM1) region (amino acids 113-134) (collectively termed ΔPrP) induce disease. A similar neurodegenerative phenotype affecting cerebellar Purkinje cells was observed in compound heterozygous Prnpo/ZHII mice and in homozygous PrnpZHII/ZHII mice (Rossi et al., 2001), whose targeted PrnpZHII allele (termed “ZHII” because it was the second targeting event of Prnp executed in Zurich) leads to overexpression of the PrPC homolog Doppel (Dpl) (Weissmann and Aguzzi, 1999). This phenotype is only observed in Prnpo/o mice that overexpress ΔPrP or Dpl and is fully reverted by substoichiometric coexpression of full-length PrPC, as either the original endogenous Prnp locus or a transgene under transcriptional control of a “half-genomic” miniversion of Prnp (Fischer et al., 1996). It follows that ΔPrP and Dpl interfere with a physiological function of PrPC.
Here we show that, in addition to CGC degeneration, mice expressing ΔPrP or Dpl in their brain suffer from widespread cerebellar leukoencephalopathy. Oligodendrocyte-specific expression of PrPC repressed this pathology and restored long-term survival but did not rescue CGC death. Conversely, neuron-specific expression of PrPC delayed disease onset and antagonized, to some extent, CGC degeneration but did not prevent white-matter degeneration. Therefore, white-matter disease and CGC degeneration are independent of each other. Surprisingly, white-matter disease appears to represent a major survival-limiting pathology induced by a toxic PrP mutant.
Materials and Methods
Generation of myelin basic protein-PrP and neuron-specific enolase-PrP transgenic mice. Mice hemizygously transgenic for the F35(PrPΔ32-134)a+/- allele (Shmerling et al., 1998) and hemizygous for the ablated endogenous Prnp locus (Prnp+/ZHI) (Büeler et al., 1992) are referred to as ΔPrPo mice. Myelin basic protein (MBP)-PrPtg/-; Prnpo/o mice (Prinz et al., 2004) were crossed to ΔPrPo or to compound PrnpZHI/ZHII-Lox1 mice (Rossi et al., 2001). The wild-type and neo-targeted Prnp loci were detected by PCR analysis of tail DNA as described previously (Büeler et al., 1992), the F35(PrPΔ32-134) transgene was detected as described previously (Shmerling et al., 1998), and the ZHII-lox1 locus was also detected as described previously (Rossi et al., 2001). To detect the MBP-PrP transgene, the following primers were used: 5′-ggc aac gtg ctg gtt att gtg c-3′ and 5′-tcc cca gca tgt agc cac caa gg-3′ (annealing at 58°C). For the generation of neuron-specific enolase (NSE)-PrP+/-; F35(PrPΔ32-134)+/- mice, NSE-PrP+/+; Prnpo/o mice were crossed with F35(PrPΔ32-134)+/-;Prnp+/0 animals. Primers used were as follows: Prnpint2, 5′-ATA CTG GGC ACT GAT ACC TTG TTC CTCA T; P10rev, 5′-GCT GGG CTT GTT CCA CTG ATT ATG GGT AC. These primers amplify selectively the wild-type Prnp allele but not the NSE-PrP transgene and were used to identify animals negative for wild-type Prnp. Standard PCR for Prnp (Büeler et al., 1992) detected both the NSE-PrP and the wild-type Prnp allele and was used to confirm the inheritance of the NSE-PrP in animals lacking wild-type Prnp. Onset of disease was declared when animals exhibited unambiguous tremor or unsteady gait that was blindly recognizable from healthy littermates in the same cage by two different observers. Animals were considered to be at a terminal disease stage when they developed hindlimb paralysis and were unable to reposition for 10 s after having been turned back-down.
Histopathology. Organs were fixed in 4% paraformaldehyde (PFA) in PBS, pH 7.5, paraffin embedded, and cut into 2 μm sections. Brain sections were stained with hematoxylin-eosin, Luxol-Nissl (myelin and neurons), and commercial antibodies to GFAP (Dako, Carpinteria, CA), microtubule-associated protein-2 (MAP-2) (Chemicon, Temecula, CA), CNPase (Roche, Basel, Switzerland), or phosphorylated neurofilament protein (Sigma, St. Louis, MO). For major histocompatibility complex class II (MHC-II) and CD11b stains, tissues were snap frozen in OCT, cut on a vibratome, and stained with antibodies to CD11b (PharMingen, San Diego, CA) or MHC-II (Serotec, Oxford, UK). Antibody stainings were visualized using the peroxidase-anti-peroxidase method or a fluorescent secondary antibody. Apoptosis was assessed using the in situ cell death kit (Roche) according to the instructions of the manufacturer.
Semithin sections and electron microscopy. Mice were perfused with ice-cold 4% PFA/3% glutaraldehyde. Brains and spinal cord were removed, immersed in the same solutions, and kept at 4°C until processing. Tissues were embedded in Epon, and semithin sections were stained with toluidine blue or used for electron microscopy.
Myelin preparation. Myelin was purified from mouse brains as described previously (Norton and Poduslo, 1973). Brain homogenates (5% w/v) were prepared in 0.32 m sucrose. An aliquot of the homogenate was stored at -80°C. The rest was layered over 0.85 m sucrose and centrifuged at 75,000 × g for 30 min. Crude myelin was collected from the interface, resuspended in water, and centrifuged at 75,000 × g for 15 min. The resultant pellet was osmotically shocked in deionized water to remove contaminants from within myelin vesicles and centrifuged twice at 12,000 × g for 15 min. The pellets were resuspended in 0.32 m sucrose, layered over 0.85 m sucrose, and centrifuged at 75,000 × g for 30 min. Myelin was again collected from the interface, resuspended in deionized water, and pelleted at 75,000 × g for 15 min to remove residual sucrose. Finally, myelin was resuspended in water and stored at -80°C. Equal amounts of myelin preparation were then processed for immunoblotting using an anti-PrP antibody [ICSM-18 (White et al., 2003) or POM-1 (M. Polymenidou and A. Aquzzi, unpublished observation)], an anti-NCAM antibody (Research Diagnostics, Flanders, NJ), or an anti-GFAP antibody (Dako, Glostrup, Denmark).
Oligodendrocyte cultures. Mixed glial cell cultures containing oligodendrocytes and astrocytes were produced from 1-d-old neonatal mice as described previously (McCarthy and de Vellis, 1980; Trotter and Schachner, 1989). Cultures were prepared with high-glucose DMEM supplemented with 10% fetal bovine serum and replenished on day 4 and every 3-4 d thereafter for 10 d with DMEM plus 10% heat-inactivated horse serum. Oligodendrocytes were purified from mixed glial cultures by differential detachment and negative selection of microglia by adherence to hydrophobic plastic. Purified oligodendrocytes were then plated onto glass or plastic culture chambers coated, respectively, with 100 or 10 μg/ml poly-l-lysine. Oligodendrocyte precursors were expanded with PDGF- and FGF-supplemented SATO medium [400 ng/ml T3, 400 ng/ml T4, 2 mmol/L glutamine, 50 U/ml penicillin, 50 g/ml streptomycin, and 5 ml of N2 supplement (Invitrogen, Paisley, UK)] for 2 d and subsequently differentiated with 1% horse serum-supplemented SATO medium. After 3 d of differentiation, cells were washed with PBS and homogenized in PBS containing 0.5% NP-40, 0.5% sodium deoxycholate, and protease inhibitors (Roche). Homogenate was then incubated on ice for 30 min and centrifuged at 10,000 rpm for 15 min at 4°C. Equal amounts of lysate were then processed for immunoblotting with a mouse monoclonal antibody to PrP (ICSM-18) (White et al., 2003) and with a mouse monoclonal anti-actin antibody (Chemicon).
Flotation assays. Flotation of detergent-insoluble complexes was performed as described previously (Naslavsky et al., 1997). Appropriate brain homogenates were extracted for 1 h on ice in cold lysis buffer (150 mm NaCl, 25 mm Tris-HCl, pH 7.5, 5 mm EDTA, and 1% Triton X-100; total protein, 30 μg in 100 μl). Extracts were mixed with two volumes (200 μl) of 60% Optiprep (Axis Shield, Oslo, Norway) to reach a final concentration of 40%. All lysates were loaded at the bottom of Beckman Instruments (Fullerton, CA) ultracentrifuge tubes. A 5-30% Optiprep step gradient in TNE (in mm: 150 NaCl, 25 Tris-HCL, pH 7.5, and 5 EDTA) was then overlaid onto the lysate (1300 μl of 30% Optiprep and 300 μl of 5% Optiprep). Tubes were centrifuged for 12 h at 4°C in a TLS55 Beckman Instruments rotor at 100,000 × g. Fractions (200 μl) were collected from the top of the tube and processed for immunoblotting and visualization with an anti-PrP antibody [ICSM-18 (White et al., 2003) or POM-1 (Polymenidou and Aquzzi, unpublished observation); both detect the PrP core and display identical patterns] and an anti-flotillin 1 antibody (BD Transduction Laboratories, Lexington, KY).
Western blot analysis of myelin proteins. Brains and spinal cords were removed and homogenized in lysis buffer containing 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% SDS, 1 mm sodium orthovanadate (Fluka, Basel, Switzerland), and protease inhibitors (Roche) in PBS. Homogenates were incubated on ice for 30 min and centrifuged at 10,000 rpm for 10 min at 4°C. Equal amounts of lysate (10 μg of protein for spinal cord and 20 μg of protein for brain homogenate) were separated by electrophoresis on 12% SDS-PAGE gels and transferred to nitrocellulose membranes. The membranes were probed with antibodies to MBP (Chemicon), myelin-associated glycoprotein (MAG) (Chemicon), proteolipid protein (PLP) (Chemicon), 2′,3′-cyclic nucleotide 3′-phosphodiesterase (CNP) (Abcam, Cambridge, UK), and actin (Chemicon). The amount of MBP, MAG, and actin was quantified using a VersaDoc model 4000 Imaging System (Bio-Rad, Zurich, Switzerland).
Northern blots. Brains were collected and immediately snap frozen in liquid nitrogen before storage at -80°C until required. Total RNA was isolated using Trizol (Invitrogen, Basel, Switzerland) with a T18 Basic Disperser (Ika Works, Staufen, Germany). Northern blot hybridization experiments were performed according to standard procedures (Sambrook and Russell, 2001) using 15 μg of total RNA. Random primed [α-32P]deoxyCTP probes were prepared from 20 ng of gel-purified DNA homologous to target transcripts using a Rediprime II random priming kit (Amersham Biosciences, Otelfingen, Switzerland) and purified with Nick Sephadex G50 columns (Amersham Biosciences). Prehybridization and hybridization were performed overnight at 42°C using Ultrahyb (Ambion, Austin, TX), containing 1 × 106 cpm/ml radiolabeled probe, 100 μg/ml denatured herring sperm DNA (Catalysis), and 20 μg/ml denatured yeast tRNA (Fluka). Posthybridization washes were performed according to Ultrahyb instructions. Hybridization signals were quantified using Fujifilm (Tokyo, Japan) imaging plate technology (BAS-1800II) and normalized for variations in RNA loading by subsequent probing with an excess of 18S ribosomal RNA cDNA probe (Miele et al., 1998).
Results
Myelinopathy and axonal degeneration in ΔPrPo mice
Hemizygous hg-PrPΔ(32-134)tg/- transgenic mice, which express amino proximally truncated PrPC under transcriptional control of a half-genomic Prnp minigene (Fischer et al., 1996) (henceforth termed ΔPrP+ mice) were bred to Prnpo/o mice. The resulting mice (termed ΔPrPo), hemizygous for the hg-PrPΔ(32-134) transgene and devoid of both endogenous Prnp alleles, developed CGC degeneration as described previously (Shmerling et al., 1998) (Fig. 1a,b). In addition, histological analysis of terminally sick ΔPrPo brains revealed additional coarse vacuolar degeneration of the intrapontine part of cranial nerves (Fig. 1d,e). Also, substantial areas of white matter in the cerebellum, the brainstem, as well as the anterior and lateral spinal cord columns exhibited large vacuoles, astrogliosis, and activated microglia/macrophages (Fig. 1m-r).
We also bred heterozygous and homozygous hg-PrPΔ(32-121)tg transgenic mice, which express a Prnp version with a smaller deletion than the hg-PrPΔ(32-134) transgene, to the Prnpo/o genetic background. We found that both the hg-PrPΔ(32-121) and hg-PrPΔ(32-134) transgenes induced analogous white-matter lesions in Prnpo/o mice (data not shown). Therefore, in all experiments described below, we used ΔPrPo mice as a general paradigm for the toxicity of amino proximally truncated PrPC.
Myelinated fibers in spinal cord and cerebellar white matter displayed severe axonal loss associated with large vacuoles and degeneration of myelin sheaths into condensed spheroids (Fig. 2a-f). Semithin sections and electron microscopy did not evidence any abnormality in myelin thickness, spacing, and periodicity in ΔPrPo and ΔPrP+ mice (Fig. 2i), but many individual sheaths displayed vacuoles, especially around large axons. Axonal pathology was evident in the latter areas. Some axons were swollen and featured accumulation of mitochondria and dense bodies (Fig. 2j,k). Other axons suffered from compromised adhesion to their myelin sheaths, evident as periaxonal splitting over broad portions of the axolemma and detachment from the myelin wrap (Fig. 2l,m). Many axons had completely degenerated, leaving behind empty myelin rings or collapsed and condensed residual myelin ghosts (Fig. 2n,o).
Axon-myelin pathology may result from developmental myelination failure or, alternatively, from degeneration of myelinated axons after completion of normal development. This question was addressed by studying semithin spinal cord and cerebellar white-matter sections from postnatal day 25 mice, shortly after completion of the myelination process. We found no vacuolation, hypomyelination, or axonal loss at this early stage (Fig. 2g,h), suggesting that pathological changes occur after myelination and are of degenerative rather than developmental origin.
Although florid apoptosis was only found in CGCs, as described previously (Shmerling et al., 1998), abnormal accumulations of perikaryal phosphorylated neurofilament proteins were visible in degenerating CGCs, as well as in brainstem and spinal cord neurons (Fig. 3), implying that neurons other than CGCs may also be affected by ΔPrP toxicity. Additionally, this suggests that axonal damage and pathologic cytoplasmic accumulation of phosphorylated neurofilament are a primary consequence of ΔPrP expression rather than secondary to neuronal cell death. We considered whether axon-myelin degeneration might result from oligodendroglial damage. However, no apoptosis was found in the white matter of diseased animals (Fig. 3), suggesting that oligodendrocyte survival was not generally impaired by ΔPrP. Therefore, general oligodendroglial cytotoxicity is unlikely to account for white-matter pathology.
White-matter pathology in Prnpo/o mice overexpressing Dpl
The prion protein homolog Dpl bears structural resemblance to ΔPrP (Luhrs et al., 2003) and is encoded by the Prnd gene, which is situated 16 kb 3′ of Prnp. In homozygous PrnpZHII/ZHII and in compound heterozygous PrnpZHII/o mice, loxP-mediated deletion of the entire third Prnp exon induces atypical intergenic splicing patterns that place Dpl under transcriptional control of the Prnp promoter, thereby leading to Dpl overexpression in brain (Weissmann and Aguzzi, 1999; Rossi et al., 2001). Overexpression of Prnd leads to cerebellar degeneration, possibly by similar mechanisms as ΔPrP (Moore et al., 1999; Genoud et al., 2004). We therefore analyzed the white matter of 18-month-old PrnpZHII/o mice with advanced clinical disease. We found white-matter vacuolation similar to that of ΔPrPo mice (Fig. 1c,f,i). Unexpectedly, white-matter disease was not restricted to cerebellum, brainstem, and spinal cord, but extended into the forebrain white matter, the pyramidal projections, and the corpus callosum (Fig. 1j-l).
Oligodendroglial PrPC represses leukoencephalopathy but not granule cell degeneration
Axons and their myelin sheaths are functionally dependent on each other: degeneration in one compartment may result in secondary damage to the other compartment. Alternatively, damage may result from parallel primary toxicity to both compartments. If myelin-producing oligodendrocytes were a primary target of mutant PrPC and Dpl, selective oligodendrocytic expression of PrPC should repress white-matter pathology. We therefore crossed ΔPrPo mice to tg640 mice, which express full-length PrPC under transcriptional control of a myelin basic protein promoter fragment (Fig. 4a) (Prinz et al., 2004).
Hemizygous expression of the MBP-PrP transgene significantly delayed the onset of tremor and ataxia (mean time of onset, 74 ± 3 vs 43 ± 2 d; p < 0.0001) (Table 1, Fig. 4a,c) and reduced the typical wasting of ΔPrPo mice (body weight at 8 weeks of age, 22.2 ± 0.39 vs 14.8+0.29 g; p < 0.0001). ΔPrPo mice reached terminal disease at 3-4 months of age, with severe wasting and hindlimb paralysis. In contrast, ΔPrPoMBP-PrPtg/- littermates were ataxic, but weight loss was not as severe as in ΔPrP° mice. The weight of 1-year-old ΔPrPoMBP-PrPtg/- was still 17-19 g. Remarkably, these mice achieved long-term survival (>400 d), with stable or minimally progressing disease, whereas the mean survival time of ΔPrPo littermates was 99 ± 5.3 d (Fig. 4a,c). Histological analysis showed repression, but not complete rescue, of axon-myelin degeneration and white-matter vacuolation (Fig. 4g-j). Instead, no rescue of CGC was observed (Fig. 4e,f), confirming that white-matter disease is unrelated to CGC degeneration. These results strengthen the conjecture that myelin is a direct target of ΔPrP toxicity.
Compound heterozygous PrnpZHII/ZHI mice develop neurological symptoms at 13-21 months and survive 18-24 months (Rossi et al., 2001). Also in these mice, intercross with tg640 mice delayed onset of disease, prolonged survival (Table 1, Fig. 4b,d), and repressed axon-myelin degeneration (Fig. 4m-p), adding support to the hypothesis that Dpl and ΔPrP share the same pathological basis. However, expression of the MBP-PrP transgene did not rescue Purkinje cell degeneration (Fig. 4k,l).
Neuronal PrPC expression antagonizes CGC degeneration but does not prevent leukoencephalopathy
Neuronal processes may conceivably represent the primary target of mutant PrP toxicity, whereas myelinopathy may be secondary to axonal degeneration. If so, oligodendrocytes may provide PrPC to the axon-myelin interface, which would abrogate any cell-autonomous axonal ΔPrP toxicity in trans. Intercellular transfer of glycosylphosphatidylinositol (GPI)-anchored protein (a phenomenon called “cell painting”) was described for CD4 (Anderson et al., 1996) and for PrPC in vitro (Liu et al., 2002). Alternatively, PrPC may act as a ligand and trigger signaling through receptors expressed by neighboring cells.
These scenarios were tested by crossing ΔPrPo mice with mice expressing murine PrPC under transcriptional control of the NSE promoter (Fig. 5a). NSE-PrP were crossed to mice lacking Prnp, thus yielding NSE-PrPtg/-Prnpo/o mice that are hemizygous for the NSE-PrP transgene and homozygous for the ZHI Prnp null allele. The latter mice (henceforth termed NSE-PrPo) express PrPC in neurons including CGC at higher levels than wild-type mice and do not develop any spontaneous pathological phenotype (O. T. Giger, B. Navarro, and A. Aguzzi, unpublished data). Instead, ΔPrPoNSE-PrP mice developed tremor and ataxia only at 9-18 months and survived >500 d. This confirms that neuronal PrPC expression prolongs survival of ΔPrPo mice (Table 1, Fig. 5b). However, histological analysis of clinically sick animals revealed leukoencephalopathy of similar severity in ΔPrPo and ΔPrPoNSE-PrP mice but not in age-matched ΔPrP+ animals (Fig. 5c-h). Leukoencephalopathy was found also in pyramidal tracts, corpus callosum, and optic tracts (data not shown), suggesting that neuronal protection, by prolonging survival, allowed for more extensive development of myelin pathology.
These findings indicate that neuron-restricted expression, even at supraphysiological levels, does not suffice to prevent ΔPrP-induced leukoencephalopathy and adds strength to the contention that ΔPrPo leukoencephalopathy is a primary myelin pathology. ΔPrPoNSE-PrP mice suffered also from CGC degeneration, although to a lesser extent than ΔPrPo mice. This suggests that neuron-restricted expression of PrPC is beneficial but insufficient to completely prevent neuronal death (Fig. 5c,d) and raises the possibility that CGC may suffer (at least in part) from deafferentation after degeneration of white-matter fibers innervating the granule cell layer.
PrPC is present in purified myelin of wild-type and MBP-PrP mice
Myelin was isolated from brains of NSE-PrPo, MBP-PrPo, Prnpo/o, ΔPrPo, and Prnp+/+ (wild-type) mice (age, 2-3 months). Sizable amounts of PrPC were detected in myelin preparations of Prnp+/+, MBP-PrP, and ΔPrPo mice (Fig. 6a). However, we found PrPC also in purified myelin of NSE-PrP mice, whose oligodendrocytes do not express PrPC (Fig. 6a,d). Poor fractionation with neuronal or astrocytic components contaminating the myelin preparations appears to be very unlikely, because GFAP and NCAM, which would be indicative of contamination, were essentially removed (Fig. 6b,c). The 120 kDa isoform of NCAM is the major isoform found in oligodendrocytes and myelin sheaths (Bhat and Silberberg, 1988). Intriguingly, ΔPrP brain homogenates displayed prominent NCAM 180, as well as two additional NCAM bands of unknown origin, which will be characterized in future studies. The presence of PrPC in myelin of NSE-PrP mice, therefore, suggests that axolemmal PrP undergoes high-affinity interactions with myelin membrane proteins, as shown for other proteins of the myelin-axolemmal complex (Menon et al., 2003). Alternatively, neuronally synthesized PrPC may be transferred onto myelin sheaths by cell painting, which was shown to be highly efficient in the case of GPI-linked proteins (Anderson et al., 1996).
Unglycosylated PrPC seems to be predominant in myelinic PrPC (Fig. 6a), similarly to what was observed in MBP-PrP mice (Prinz et al., 2004). The intense low-molecular-weight PrPC bands in MBP-PrP mice might reflect degradation of unglycosylated PrPC, because glycosylation is known to influence the stability of proteins.
ΔPrP and PrPC in MBP-PrP mice are localized in lipid rafts
PrPC localizes to lipid rafts (Naslavsky et al., 1997; Meier et al., 2003), in which it may exert a physiological function by interacting with signaling molecules. Dpl is also found in rafts (Massimino et al., 2004). If toxicity of ΔPrP and Dpl results from interference with some PrPC-mediated signaling pathway and toxicity is counteracted by PrPC, these three proteins may compete for the same partners on rafts. Moreover, ΔPrP and Dpl may disrupt normal axonglia interaction and adhesion, whose integrity depends on GPI-anchored and transmembrane proteins (Falk et al., 2002; Marcus et al., 2002; Bartsch, 2003; Vinson et al., 2003). To investigate the topology of ΔPrP, we performed flotation assays on Optiprep gradients with cold Triton X-100-treated ΔPrPo brain extracts. ΔPrP was detected in low-density, detergent-insoluble fractions and displayed buoyancy similar to that of PrP and flotillin (Fig. 6f-j). These characteristics strongly suggest that ΔPrP is indeed localized in lipid rafts and colocalized with wild-type PrPC. Therefore, cytotoxicity is unlikely to be attributable to inappropriate membrane targeting of ΔPrP. In addition, PrPC from MBP-PrP mice associated with rafts similarly to PrPC from wild-type mice (Fig. 6i). Unglycosylated PrPC was the dominant moiety, similar to PrPC isolated from myelin.
Neither ΔPrP nor Dpl affect myelin protein expression
Leukoencephalopathy in ΔPrPo and PrnpZHII/o mice had features suggestive of compromised axonglial adhesion similar to those described in mice lacking MAG and myelin galactolipids (Marcus et al., 2002), as well as in mice deficient for PLP (Griffiths et al., 1998). In addition, we found axonal swellings similar to those of Cnp1-/- mice (Lappe-Siefke et al., 2003). Moreover, in mice deficient for these diverse myelin proteins, oligodendrocytes produce appropriately thick myelin sheaths. These findings raised the question as to whether ΔPrP and Dpl might affect the synthesis or the stability of the above myelin components. Therefore, we analyzed the myelin content in whole brain and spinal cord homogenates of Prnpo/o, Prnp+/+, ΔPrPo, and PrnpZHII/o mice (age, 2 months). MAG, MBP, PLP, and CNP transcription and protein content were assayed with Northern and Western blots (Fig. 7a,b). The amount of MAG and MBP protein was normalized against actin (Fig. 7c). These analyses revealed that both the transcription and the steady-state protein concentration of all investigated myelin proteins were unchanged in ΔPrPo and Prn-pZHII/o mice.
Discussion
Overexpression of amino proximally truncated PrPC variants collectively termed ΔPrP, (Shmerling et al., 1998), or of the structural PrPC homolog Dpl (Weissmann and Aguzzi, 1999; Behrens et al., 2002), triggers neuronal degeneration. We now report that both ΔPrP and Dpl elicit additional leukoencephalopathy. These pathologies were not entirely expected, because PrP-related diseases are thought to affect primarily neurons (Chiesa et al., 1998, 2000; Hegde et al., 1998, 1999; Shmerling et al., 1998; Rossi et al., 2001). Conversely, myelin sheath damage and neuroaxonal dystrophy (Liberski et al., 2002), as well as cytopathic changes in oligodendrocytes (El Hachimi et al., 1998), have been noted in human and animal TSEs.
Myelinopathy, rather than neuronal damage, may be a major determinant of lethality in Shmerling's syndrome, because oligodendrocyte-restricted expression of PrPC selectively suppresses leukoencephalopathy and restores long-term survival. Instead, neuron-restricted expression of PrPC, even at supraphysiological levels, does not prevent white-matter degeneration. These data indicate that leukoencephalopathy is not just a consequence of neurodegeneration but suggest a primary role of oligodendrocytes in white-matter disease. Oligodendrocyte-restricted PrPC may repress in trans the axonal pathology elicited by mutant PrP. A cellular signal might be transduced through PrPC from oligodendrocytes to axons, analogously to other GPI-anchored proteins (such as F3 and NCAM 120) that play a role in axon-myelin interactions (Falk et al., 2002). Because PrPC binds in vitro to the transmembrane and GPI-linked forms of NCAM (Schmitt-Ulms et al., 2001), it might contribute to the very same pathways. Alternatively, PrPC may instruct oligodendrocytes to release factors necessary for axonal survival (Wilkins et al., 2003). Direct toxicity of ΔPrP and Dpl to oligodendrocytes seems to be unlikely, because we did not observe oligodendrocytic cell death or ultrastructural abnormalities in oligodendrocytes. Both oligodendrocyte- and neuron-restricted expression delayed the clinical course of Shmerling's disease, yet only endogenous PrPC expression, which occurs on both neurons and glia (Moser et al., 1995), brought about complete rescue. Therefore, both neuronal and oligodendrocyte PrPC expression are needed for maintenance of the axon-myelin interface in ΔPrP mice.
Surprisingly, neuron-specific PrPC expression, even at supraphysiological levels, did not completely rescue CGC degeneration. Perhaps ongoing degeneration of white-matter fibers innervating the granule cell layer participates indirectly in granule cell degeneration, e.g., by deafferentation and neuronal “dying back.” The delayed onset of myelinopathy in ΔPrPoNSE-PrP mice might be related to the presence of PrPC in NSE-PrP myelin which, in turn, might be attributable to GPI cell painting or to high-affinity interaction of axonal PrPC with myelin components.
Leukoencephalopathy in ΔPrPo and PrnpZHII/o mice had features suggestive of compromised axonglial adhesion, including axonal swellings similar to those described in mice lacking MAG and myelin galactolipids (Marcus et al., 2002), PLP (Griffiths et al., 1998), and Cnp1 (Lappe-Siefke et al., 2003). However, expression of MAG, MBP, PLP, and CNP was not affected in ΔPrPo and PrnpZHII/o mice, suggesting that disturbance of myelin and axonal physiology is directly mediated by ΔPrP and Dpl rather than by dysmetabolism of myelin proteins.
The overlapping patterns of (1) Dpl- and ΔPrP-induced leukoencephalopathy, (2) their modulation by coexpression of full-length PrPC (Moore et al., 1999; Li et al., 2000; Rossi et al., 2001), (3) the structural homology between Dpl and ΔPrP (Luhrs et al., 2003), and (4) the abrogation of neurodegeneration by removal of Dpl from PrnpZHII/ZHII mice (Genoud et al., 2004) all suggest that ΔPrP and Dpl trigger similar pathogenetic mechanisms. What might be the common structural basis of such pathologies? Of all amino proximal deletions that were tested (Shmerling et al., 1998), only those that disrupt the highly conserved hydrophobic TM1 region (amino acids 113-134) provoked pathologies that were prevented by wild-type PrPC.
Two possible molecular mechanism of toxicity have been put forward. Shmerling et al. (1998) proposed that the amino proximal region of PrP represents an effector domain, whose deletion abolishes trophic signaling and partially reduces the affinity of PrPC for its hypothetical ligands. According to this scenario, removal of the amino proximal confers recessive-negative toxic properties to ΔPrP that are abolished if ΔPrP is displaced by PrPC. In wild-type mice, oligodendrocytic PrPC might transduce survival signals by interacting with axonal molecules. Leukoencephalopathy may occur because ΔPrP abrogates such trophic signals. The absence of such phenotype in Prnp-/- mice suggests the existence of a functional homolog provisionally termed π (Shmerling et al., 1998).
Alternatively, PrPC and its homologs may form homo- and hetero-oligomers. Although ΔPrP and Dpl homo-oligomers may be toxic to neurons, participation of full-length PrPC to such complexes would “dope” their stoichiometry, hence abolishing their toxicity (Behrens and Aguzzi, 2002).
It will be interesting to test which of the above hypotheses, if any, reflects the mechanism of ΔPrP/Dpl-associated damage to the axon-myelin complex. The finding that full-length prion protein, specifically expressed in oligodendrocytes and found in myelin, represses these defects suggests that PrPC plays a role in myelin physiology, oligodendroglial-mediated axonal support, and axon-myelin interactions. This contention is supported by the finding that aged Prnp-/- mice suffer from peripheral demyelination. Both leukoencephalopathy and CGC degeneration are suppressed by coexpression of PrPC, suggesting that the molecular pathogenesis of these two diseases is similar.
In addition to its role in prion disease pathogenesis, repression of ΔPrP/Dpl-associated neurodegenerative pathologies is likely to represent the most thoroughly characterized function of PrPC. As shown by the present study, the neurodegenerative syndromes induced by ΔPrP provide an accessible experimental system for studying function of PrPC in vivo by reverse genetics.
Footnotes
This work was supported by grants from the Bundesamt für Bildung und Wissenschaft (European Union), the Swiss National Research Foundation, the United States National Prion Research Program, and the National Centre of Competence in Research on Neural Plasticity and Repair (A.A.). I.R. and N.B. received MD-PhD fellowships from the Swiss Cancer League and the Swiss National Research Foundation. We thank Petra Schwarz for skillful and dedicated help throughout this project and Markus Glatzel for help with NSE-PrP transgenic mice.
Correspondence should be addressed to Dr. Adriano Aguzzi, Institute of Neuropathology, University Hospital of Zurich, Schmelzbergstrasse 12, CH-8091 Zurich, Switzerland. E-mail: adriano{at}pathol.unizh.ch.
I. Radovanovic's present address: Department of Neurosurgery, University Hospital of Geneva, 24 rue Micheli-du-Crest, CH-1211 Geneva, Switzerland.
M. Prinz's present address: Institute of Neuropathology, Georg-August-University Göttingen, Robert-Koch-Strasse 40, D-37075 Göttingen, Germany.
B. Navarro's present address: Centro sui virus e le virosi delle colture Mediterranee, Università degli Studi di Bari, Via Amendola 165/A, 70126 Bari, Italy.
Copyright © 2005 Society for Neuroscience 0270-6474/05/254879-10$15.00/0
↵* I.R. and N.B. contributed equally to this work.